PCO 2014 in conjunction with IEEE IPDPS 2014
Special Session on GPU Computing of PDP 2014
iThings 2013 (CCTV report)
The CDA team plays a part in addressing the different challenges in distributed computing. We consider both traditional domains of research related to the design of distributed algorithms like determining the global state of a distributed system for distributed determination detection purpose and new challenges related to the solution of complex or large scale problems in combinatorial optimization or numerical simulation using techniques such as peer to peer distributed computing in a massive parallelism context. As a consequence, the research work of the team concerns as well the design and analysis of distributed algorithms as the design of distributed computing environments. The former topic is considered for example in the framework of ANR project Smart Surface
, the latter topic is studied in the framework of the ANR project CIP
CDA team works in particular on hybrid computing (see Recent publications
). Dr Didier El Baz has received recently an NVIDIA Academic Partnership.
The CDA team aims particularly at developing the concept of asynchronism which constitutes a major challenge for the advancement of distributed computing at a very large scale. This concept is gaining considerable attention today since large distributed systems must cope with the very asynchronous nature of communication networks, dynamicity and the presence of faults in the system. In particular, researchs are made on asynchronous iterative algorithms which are particularly well suited to the massive distribution context. Studies are developed for high performance distributed processing on dedicated architectures as in the ANR Smart Surface
project and on general architectures.